Vinod Namboodiri and Lixin Gao

1
Prediction-Based Routing for Vehicular Ad Hoc
Networks

• Vinod Namboodiri and Lixin Gao
• University of Massachusetts Amherst
• IEEE Transactions on Vehicular Technology, July
  2007
• speaker Yu-Hsun Chen

2
Outline

• Introduction
• Related Work
• Highway Mobility Model
• PBR Protocol
• Experiments and Results
• Conclusion

3
Introduction 1

• Connectivity while on the road will be an
  important application area
• Gaming and multimedia streaming
• Safety application
• Low cost on safety products
• Considerations to achieve connectivity while on
  the road
• Bandwidth
• Cost
• seamless mobility

4
Introduction 2

• Wireless technology for Internet access
• 3G, 4G, WiMax, satellite-based
• seamless connectivity
• More expansive
• Wireless LAN
• Low-cost
• High bandwidth
• Capability of ad hoc mode
• Limited range

5
Introduction 3

• Wireless connectivity from a vehicle
• Inter-vehicular communications (IVCs)
• Internet connectivity
• Static gateways alongside roads 34
• Deployment cost
• Route switching between gateways
• Mobility gateway approach
• Low cost
• Without geopolitical boundaries
• Fewer gateway switches

6
Introduction 4
7
Introduction 5

• Contributions
• Mobility model
• Highway mobility patterns
• Prediction-based routing protocol
• Predict how long routes will last
• Preemptively creates new routes to replace old
  ones before they break

8
Outline

• Introduction
• Related Work
• Highway Mobility Model
• PBR Protocol
• Experiments and Results
• Conclusion

9
Mobile Ad hoc Routing

• Proactive
• All nodes send routing messages at predetermined
  periods
• Difficulty what messaging period is best to
  maximize routing performance
• Reactive
• On-demand basis
• Lack sensitivity toward new better routes
• Location-based
• Rely on a location server
• Overhead to maintain vehicles current information

10
VANET Routing Classification
J. Bernsen and D. Manivannan, Unicast routing
protocols for vehicular ad hoc networks A
critical comparison and classification,
Pervasive and Mobile Computing, 2009
11
Outline

• Introduction
• Related Work
• Highway Mobility Model
• PBR Protocol
• Experiments and Results
• Conclusion

12
Highway Mobility Model 1

• Assumption all vehicles are within certain speed
  bounds
• Discrete time model
• Car speed

13
Highway Mobility Model 2

14
Outline

• Introduction
• Related Work
• Highway Mobility Model
• PBR Protocol
• Experiments and Results
• Conclusion

15
Prediction-Based Routing Protocol

• Obtaining location and velocity information of
  vehicles on the route to the gateway
• Prediction algorithm uses this information to
  predict when the route will break

16
Basic Operation 1

• When a node needs to communicate to a Internet
  location
• Broadcast an RREQ
• TTL, sequence number, source ID, destination ID,
  source nodes direction, a list of nodes and
  their directions
• A neighbor receiving the RREQ forwards it if
• 1. TTL gt 1 and higher sequence number
• 2. TTL gt 1, the same sequence number as previous
  packet, and all intermediate nodes traveling in
  the same direction

17
Basic Operation 2

• When the RREQ reaches a gateway with the desired
  route to the sought destination
• The gateway send back an RREP using the chain of
  nodes in the RREQ
• When multiple gateways reply
• 1. choose the gateway with minimum hops and all
  nodes on the route are moving in the same
  direction as itself
• 2. choose the gateway with minimum hops
• When multiple route from the same gateway
• Choose the route that has the maximum predicted
  route lifetime

18
Basic Operation 3

• The RREP in conjunction with the prediction
  algorithm is used to give the source a predicted
  lifetime for the route
• The source sends out a new RREQ just before this
  timer expires
• The preemption interval is adaptive based on the
  lifetime of the route
• If the last packet was sent before a certain time
  threshold pred-timeout
• Turn off the preemptive route creation procedure

19
Obtaining Route Lifetime

• Information in the RREP
• Location and velocity information
• Set a lifetime field in the RREP
• Gateway lifetime ? maxlifetime
• Intermediate node predict the life time using
  the prediction algorithm
• If the lifetime value is smaller than the
  lifetime mentioned in the RREP packet
• Replace the lifetime field in the RREP

20
Prediction Algorithm 1

21
Prediction Algorithm 2
The lifetime for the route ACDE is 10s
22
Moving Closer Condition
bonus
23
Link on Oncoming Traffic
24
PBR Variants

• PBR
• A new route is constructed to the nearest gateway
• PBR-S
• Stick with a gateway as much as possible to avoid
  gateway switching
• PBR-M
• Select the gateway among all those within a
  certain number of hops with the largest predicted
  route lifetime

25
Outline

• Introduction
• Related Work
• Highway Mobility Model
• PBR Protocol
• Experiments and Results
• Conclusion

26
Routing Metrics and Simulation Environment

• Routing Metrics
• Packet delivery ratio
• Route failures (percentage of dropped packets)
• Number of RREQs

27
Effect of Vehicle and Gateway Density on
Connectivity
28
Effect of Node and Gateway Density on Routing
Performance
The number of gateways fix at 10
29
Effect of Node and Gateway Density on Routing
Performance
The number of nodes fix at 50
30
Effect of Mobility Pattern on Routing Performance
of nodes, of gateways
31
Comparing PBR with Reactive and Proactive
Protocols
10 gateways 40 nodes
32
Minimizing Gateway Switching
10 gateways
33
Conclusion

• The predictable motion of vehicles could be
  exploited to predict route failures
• A PBR protocol is presented
• Reduction in route failure
• Higher packet delivery ratio
• Keeping control overhead in check